From c94cae53f9e564484f906a79be5639fc66e8cb02 Mon Sep 17 00:00:00 2001 From: Eric Trudeau Date: Wed, 4 Dec 2013 11:39:33 +0000 Subject: XEN: Grant table address, xen_hvm_resume_frames, is a phys_addr not a pfn From: Eric Trudeau xen_hvm_resume_frames stores the physical address of the grant table. englighten.c was incorrectly setting it as if it was a page frame number. This caused the table to be mapped into the guest at an unexpected physical address. Additionally, a warning is improved to include the grant table address which failed in xen_remap. Signed-off-by: Eric Trudeau Signed-off-by: Stefano Stabellini diff --git a/arch/arm/xen/enlighten.c b/arch/arm/xen/enlighten.c index 83e4f95..6a288c7 100644 --- a/arch/arm/xen/enlighten.c +++ b/arch/arm/xen/enlighten.c @@ -224,10 +224,10 @@ static int __init xen_guest_init(void) } if (of_address_to_resource(node, GRANT_TABLE_PHYSADDR, &res)) return 0; - xen_hvm_resume_frames = res.start >> PAGE_SHIFT; + xen_hvm_resume_frames = res.start; xen_events_irq = irq_of_parse_and_map(node, 0); pr_info("Xen %s support found, events_irq=%d gnttab_frame_pfn=%lx\n", - version, xen_events_irq, xen_hvm_resume_frames); + version, xen_events_irq, (xen_hvm_resume_frames >> PAGE_SHIFT)); xen_domain_type = XEN_HVM_DOMAIN; xen_setup_features(); diff --git a/drivers/xen/grant-table.c b/drivers/xen/grant-table.c index 02838719..aa846a4 100644 --- a/drivers/xen/grant-table.c +++ b/drivers/xen/grant-table.c @@ -1176,7 +1176,8 @@ static int gnttab_setup(void) gnttab_shared.addr = xen_remap(xen_hvm_resume_frames, PAGE_SIZE * max_nr_gframes); if (gnttab_shared.addr == NULL) { - pr_warn("Failed to ioremap gnttab share frames!\n"); + pr_warn("Failed to ioremap gnttab share frames (addr=0x%08lx)!\n", + xen_hvm_resume_frames); return -ENOMEM; } } -- cgit v0.10.2 From b6497b383f65dd1bc60ea1f5d70e157a268fbd15 Mon Sep 17 00:00:00 2001 From: Ian Campbell Date: Fri, 6 Dec 2013 17:55:56 +0000 Subject: xen: privcmd: do not return pages which we have failed to unmap This failure represents a hypervisor issue, but if it does occur then nothing good can come of returning pages which still refer to a foreign owned page into the general allocation pool. Instead we are forced to leak them. Log that we have done so. The potential for failure only exists for autotranslated guest (e.g. ARM and x86 PVH). Signed-off-by: Ian Campbell Signed-off-by: Stefano Stabellini Reviewed-by: David Vrabel diff --git a/drivers/xen/privcmd.c b/drivers/xen/privcmd.c index 8e74590..569a13b 100644 --- a/drivers/xen/privcmd.c +++ b/drivers/xen/privcmd.c @@ -533,12 +533,17 @@ static void privcmd_close(struct vm_area_struct *vma) { struct page **pages = vma->vm_private_data; int numpgs = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; + int rc; if (!xen_feature(XENFEAT_auto_translated_physmap) || !numpgs || !pages) return; - xen_unmap_domain_mfn_range(vma, numpgs, pages); - free_xenballooned_pages(numpgs, pages); + rc = xen_unmap_domain_mfn_range(vma, numpgs, pages); + if (rc == 0) + free_xenballooned_pages(numpgs, pages); + else + pr_crit("unable to unmap MFN range: leaking %d pages. rc=%d\n", + numpgs, rc); kfree(pages); } -- cgit v0.10.2 From 02ab71cdae248533620abefa1d46097581457110 Mon Sep 17 00:00:00 2001 From: Stefano Stabellini Date: Mon, 9 Dec 2013 15:55:11 +0000 Subject: xen/arm64: do not call the swiotlb functions twice On arm64 the dma_map_ops implementation is based on the swiotlb. swiotlb-xen, used by default in dom0 on Xen, is also based on the swiotlb. Avoid calling into the default arm64 dma_map_ops functions from xen_dma_map_page, xen_dma_unmap_page, xen_dma_sync_single_for_cpu, and xen_dma_sync_single_for_device otherwise we end up calling into the swiotlb twice. When arm64 gets a non-swiotlb based implementation of dma_map_ops, we'll probably have to reintroduce dma_map_ops calls in page-coherent.h. Signed-off-by: Stefano Stabellini CC: catalin.marinas@arm.com CC: Will.Deacon@arm.com CC: Ian.Campbell@citrix.com diff --git a/arch/arm64/include/asm/xen/page-coherent.h b/arch/arm64/include/asm/xen/page-coherent.h index 2820f1a..dde3fc9 100644 --- a/arch/arm64/include/asm/xen/page-coherent.h +++ b/arch/arm64/include/asm/xen/page-coherent.h @@ -23,25 +23,21 @@ static inline void xen_dma_map_page(struct device *hwdev, struct page *page, unsigned long offset, size_t size, enum dma_data_direction dir, struct dma_attrs *attrs) { - __generic_dma_ops(hwdev)->map_page(hwdev, page, offset, size, dir, attrs); } static inline void xen_dma_unmap_page(struct device *hwdev, dma_addr_t handle, size_t size, enum dma_data_direction dir, struct dma_attrs *attrs) { - __generic_dma_ops(hwdev)->unmap_page(hwdev, handle, size, dir, attrs); } static inline void xen_dma_sync_single_for_cpu(struct device *hwdev, dma_addr_t handle, size_t size, enum dma_data_direction dir) { - __generic_dma_ops(hwdev)->sync_single_for_cpu(hwdev, handle, size, dir); } static inline void xen_dma_sync_single_for_device(struct device *hwdev, dma_addr_t handle, size_t size, enum dma_data_direction dir) { - __generic_dma_ops(hwdev)->sync_single_for_device(hwdev, handle, size, dir); } #endif /* _ASM_ARM64_XEN_PAGE_COHERENT_H */ -- cgit v0.10.2 From a7892f32cc3534d4cc0e64b245fbf47a8e364652 Mon Sep 17 00:00:00 2001 From: Ian Campbell Date: Wed, 11 Dec 2013 17:02:27 +0000 Subject: arm: xen: foreign mapping PTEs are special. These mappings are in fact special and require special handling in privcmd, which already exists. Failure to mark the PTE as special on arm64 causes all sorts of bad PTE fun. e.g. e.g.: BUG: Bad page map in process xl pte:e0004077b33f53 pmd:4079575003 page:ffffffbce1a2f328 count:1 mapcount:-1 mapping: (null) index:0x0 page flags: 0x4000000000000014(referenced|dirty) addr:0000007fb5259000 vm_flags:040644fa anon_vma: (null) mapping:ffffffc03a6fda58 index:0 vma->vm_ops->fault: privcmd_fault+0x0/0x38 vma->vm_file->f_op->mmap: privcmd_mmap+0x0/0x2c CPU: 0 PID: 2657 Comm: xl Not tainted 3.12.0+ #102 Call trace: [] dump_backtrace+0x0/0x12c [] show_stack+0x14/0x1c [] dump_stack+0x70/0x90 [] print_bad_pte+0x12c/0x1bc [] unmap_single_vma+0x4cc/0x700 [] unmap_vmas+0x68/0xb4 [] unmap_region+0xcc/0x1d4 [] do_munmap+0x218/0x314 [] vm_munmap+0x44/0x64 [] SyS_munmap+0x24/0x34 Where unmap_single_vma contains inlined -> unmap_page_range -> zap_pud_range -> zap_pmd_range -> zap_pte_range -> print_bad_pte. Or: BUG: Bad page state in process xl pfn:4077b4d page:ffffffbce1a2f8d8 count:0 mapcount:-1 mapping: (null) index:0x0 page flags: 0x4000000000000014(referenced|dirty) Modules linked in: CPU: 0 PID: 2657 Comm: xl Tainted: G B 3.12.0+ #102 Call trace: [] dump_backtrace+0x0/0x12c [] show_stack+0x14/0x1c [] dump_stack+0x70/0x90 [] bad_page+0xc4/0x110 [] free_pages_prepare+0xd0/0xd8 [] free_hot_cold_page+0x28/0x178 [] free_hot_cold_page_list+0x38/0x60 [] release_pages+0x190/0x1dc [] unmap_region+0x15c/0x1d4 [] do_munmap+0x218/0x314 [] vm_munmap+0x44/0x64 [] SyS_munmap+0x24/0x34 x86 already gets this correct. 32-bit arm gets away with this because there is not PTE_SPECIAL bit in the PTE there and the vm_normal_page fallback path does the right thing. Signed-off-by: Ian Campbell Signed-off-by: Stefano Stabellini diff --git a/arch/arm/xen/enlighten.c b/arch/arm/xen/enlighten.c index 6a288c7..8550123 100644 --- a/arch/arm/xen/enlighten.c +++ b/arch/arm/xen/enlighten.c @@ -96,7 +96,7 @@ static int remap_pte_fn(pte_t *ptep, pgtable_t token, unsigned long addr, struct remap_data *info = data; struct page *page = info->pages[info->index++]; unsigned long pfn = page_to_pfn(page); - pte_t pte = pfn_pte(pfn, info->prot); + pte_t pte = pte_mkspecial(pfn_pte(pfn, info->prot)); if (map_foreign_page(pfn, info->fgmfn, info->domid)) return -EFAULT; -- cgit v0.10.2 From 380108d891acf8db5cf0d477176c7ed2b62b7928 Mon Sep 17 00:00:00 2001 From: Julien Grall Date: Tue, 3 Dec 2013 15:40:37 +0000 Subject: xen/block: Correctly define structures in public headers on ARM32 and ARM64 On ARM (32 bits and 64 bits), the double-word is 8-bytes aligned. This will result on different structure from Xen and Linux repositories. As Linux is using __packed__ attribute, it must have a 4-bytes padding before each "id" field. This change breaks guest block support with older kernel. IMHO, it's acceptable because Xen on ARM is still on Tech Preview and the hypercall ABI is not yet freezed. Only one architecture (x86_32) doesn't have 64-bit ABI for the block interface. Don't add padding if Linux is compiled for this architecture. Signed-off-by: Julien Grall Acked-by: Roger Pau Monne Acked-by: David Vrabel Cc: Boris Ostrovsky Cc: Ian Campbell Acked-by: Stefano Stabellini [I had asked for confirmation that it did not break x86 and Ian went beyound the call of duty to confirm it. Also a internal regression bucket with 32/64 dom0 with 32/64 domU (PV and HVM) confirmed no regressions. ABI changes are a drag..] Signed-off-by: Konrad Rzeszutek Wilk diff --git a/include/xen/interface/io/blkif.h b/include/xen/interface/io/blkif.h index 65e1209..ae665ac 100644 --- a/include/xen/interface/io/blkif.h +++ b/include/xen/interface/io/blkif.h @@ -146,7 +146,7 @@ struct blkif_request_segment_aligned { struct blkif_request_rw { uint8_t nr_segments; /* number of segments */ blkif_vdev_t handle; /* only for read/write requests */ -#ifdef CONFIG_X86_64 +#ifndef CONFIG_X86_32 uint32_t _pad1; /* offsetof(blkif_request,u.rw.id) == 8 */ #endif uint64_t id; /* private guest value, echoed in resp */ @@ -163,7 +163,7 @@ struct blkif_request_discard { uint8_t flag; /* BLKIF_DISCARD_SECURE or zero. */ #define BLKIF_DISCARD_SECURE (1<<0) /* ignored if discard-secure=0 */ blkif_vdev_t _pad1; /* only for read/write requests */ -#ifdef CONFIG_X86_64 +#ifndef CONFIG_X86_32 uint32_t _pad2; /* offsetof(blkif_req..,u.discard.id)==8*/ #endif uint64_t id; /* private guest value, echoed in resp */ @@ -175,7 +175,7 @@ struct blkif_request_discard { struct blkif_request_other { uint8_t _pad1; blkif_vdev_t _pad2; /* only for read/write requests */ -#ifdef CONFIG_X86_64 +#ifndef CONFIG_X86_32 uint32_t _pad3; /* offsetof(blkif_req..,u.other.id)==8*/ #endif uint64_t id; /* private guest value, echoed in resp */ @@ -184,7 +184,7 @@ struct blkif_request_other { struct blkif_request_indirect { uint8_t indirect_op; uint16_t nr_segments; -#ifdef CONFIG_X86_64 +#ifndef CONFIG_X86_32 uint32_t _pad1; /* offsetof(blkif_...,u.indirect.id) == 8 */ #endif uint64_t id; @@ -192,7 +192,7 @@ struct blkif_request_indirect { blkif_vdev_t handle; uint16_t _pad2; grant_ref_t indirect_grefs[BLKIF_MAX_INDIRECT_PAGES_PER_REQUEST]; -#ifdef CONFIG_X86_64 +#ifndef CONFIG_X86_32 uint32_t _pad3; /* make it 64 byte aligned */ #else uint64_t _pad3; /* make it 64 byte aligned */ -- cgit v0.10.2 From c1d15f5c8bc1170dafe16e988e55437245966dfe Mon Sep 17 00:00:00 2001 From: Stefano Stabellini Date: Wed, 11 Dec 2013 16:58:42 +0000 Subject: xen/balloon: Seperate the auto-translate logic properly (v2) The auto-xlat logic vs the non-xlat means that we don't need to for auto-xlat guests (like PVH, HVM or ARM): - use P2M - use scratch page. However the code in increase_reservation does modify the p2m for auto_translate guests, but not in decrease_reservation. Fix that by avoiding any p2m modifications in both increase_reservation and decrease_reservation for auto_translated guests. And also avoid allocating or using scratch pages for auto_translated guests. Lastly, since !auto-xlat is really another way of saying 'xen_pv' remove the redundant 'xen_pv_domain' check. Signed-off-by: Stefano Stabellini [v2: Updated the description] Signed-off-by: Konrad Rzeszutek Wilk diff --git a/drivers/xen/balloon.c b/drivers/xen/balloon.c index 55ea73f..4c02e2b 100644 --- a/drivers/xen/balloon.c +++ b/drivers/xen/balloon.c @@ -350,17 +350,19 @@ static enum bp_state increase_reservation(unsigned long nr_pages) pfn = page_to_pfn(page); - set_phys_to_machine(pfn, frame_list[i]); - #ifdef CONFIG_XEN_HAVE_PVMMU - /* Link back into the page tables if not highmem. */ - if (xen_pv_domain() && !PageHighMem(page)) { - int ret; - ret = HYPERVISOR_update_va_mapping( - (unsigned long)__va(pfn << PAGE_SHIFT), - mfn_pte(frame_list[i], PAGE_KERNEL), - 0); - BUG_ON(ret); + if (!xen_feature(XENFEAT_auto_translated_physmap)) { + set_phys_to_machine(pfn, frame_list[i]); + + /* Link back into the page tables if not highmem. */ + if (!PageHighMem(page)) { + int ret; + ret = HYPERVISOR_update_va_mapping( + (unsigned long)__va(pfn << PAGE_SHIFT), + mfn_pte(frame_list[i], PAGE_KERNEL), + 0); + BUG_ON(ret); + } } #endif @@ -378,7 +380,6 @@ static enum bp_state decrease_reservation(unsigned long nr_pages, gfp_t gfp) enum bp_state state = BP_DONE; unsigned long pfn, i; struct page *page; - struct page *scratch_page; int ret; struct xen_memory_reservation reservation = { .address_bits = 0, @@ -411,27 +412,29 @@ static enum bp_state decrease_reservation(unsigned long nr_pages, gfp_t gfp) scrub_page(page); +#ifdef CONFIG_XEN_HAVE_PVMMU /* * Ballooned out frames are effectively replaced with * a scratch frame. Ensure direct mappings and the * p2m are consistent. */ - scratch_page = get_balloon_scratch_page(); -#ifdef CONFIG_XEN_HAVE_PVMMU - if (xen_pv_domain() && !PageHighMem(page)) { - ret = HYPERVISOR_update_va_mapping( - (unsigned long)__va(pfn << PAGE_SHIFT), - pfn_pte(page_to_pfn(scratch_page), - PAGE_KERNEL_RO), 0); - BUG_ON(ret); - } -#endif if (!xen_feature(XENFEAT_auto_translated_physmap)) { unsigned long p; + struct page *scratch_page = get_balloon_scratch_page(); + + if (!PageHighMem(page)) { + ret = HYPERVISOR_update_va_mapping( + (unsigned long)__va(pfn << PAGE_SHIFT), + pfn_pte(page_to_pfn(scratch_page), + PAGE_KERNEL_RO), 0); + BUG_ON(ret); + } p = page_to_pfn(scratch_page); __set_phys_to_machine(pfn, pfn_to_mfn(p)); + + put_balloon_scratch_page(); } - put_balloon_scratch_page(); +#endif balloon_append(pfn_to_page(pfn)); } @@ -627,15 +630,17 @@ static int __init balloon_init(void) if (!xen_domain()) return -ENODEV; - for_each_online_cpu(cpu) - { - per_cpu(balloon_scratch_page, cpu) = alloc_page(GFP_KERNEL); - if (per_cpu(balloon_scratch_page, cpu) == NULL) { - pr_warn("Failed to allocate balloon_scratch_page for cpu %d\n", cpu); - return -ENOMEM; + if (!xen_feature(XENFEAT_auto_translated_physmap)) { + for_each_online_cpu(cpu) + { + per_cpu(balloon_scratch_page, cpu) = alloc_page(GFP_KERNEL); + if (per_cpu(balloon_scratch_page, cpu) == NULL) { + pr_warn("Failed to allocate balloon_scratch_page for cpu %d\n", cpu); + return -ENOMEM; + } } + register_cpu_notifier(&balloon_cpu_notifier); } - register_cpu_notifier(&balloon_cpu_notifier); pr_info("Initialising balloon driver\n"); -- cgit v0.10.2